#! /bin/sh -eu

set -o pipefail

src=$(realpath $1)
out=$(realpath $2)

echo "Fetching $src to $out"

mkdir $out

# Configure cargo to fetch from a local copy of the crates.io registry
#
# Unfortunately, `cargo fetch` will create an output directory named after a
# hash of the registry index URL.
#
# This makes things difficult for us because we don't want our output to change
# just because the path to the registry changed, otherwise we'd have to update
# all deps' SHA256 hashes whenever we simply update the registry to a newer
# commit.
#
# Also, since cargo doesn't seem to support relative URLs in the format
# file://../path, we use a hack to make sure the registry index path/URL is
# always the same: we'll create a symlink in the current working directory to
# the real registry path, while pointing cargo to the following fixed absolute
# path:
#
# file:///proc/self/cwd/symlink-name

ln -s $rustRegistry $src/cargo-rust-registry

# TODO: replace /proc/self/cwd hack with normal relative path. Probably
# needs cargo fix.
cat <<EOF > $out/config
[registry]
index = "file:///proc/self/cwd/cargo-rust-registry"
EOF

export CARGO_HOME=$out
cd $src
cargo fetch --verbose

# TODO: check that Cargo.lock exists, and hasn't changed
# TODO: this should be done by cargo itself

# Make it deterministic

# The registry index changes all the time, so it's not deterministic
rm -rf $out/registry/index

# Make git DBs deterministic
# TODO: test with git submodules
[[ ! -d $out/git/checkouts ]] || (cd $out/git/checkouts && for name in *; do
    cd "$out/git/checkouts/$name"
    revs=""
    for branch in *; do
        cd "$branch"
        rev="$(git rev-parse HEAD)"
        revs="$revs $rev"
        cd ..
    done

    (
        # The following code was adapted from nix-prefetch-git

        cd "$out/git/db/$name"

        export GIT_DIR=.

        # Remove all remote branches
        git branch -r | while read branch; do
            git branch -rD "$branch" >&2
        done

        # Remove tags that don't point to any HEAD
        git tag | while read tag; do
            rev="$(git rev-parse $tag)"
            if [[ $revs != *" $rev"* ]]; then
                git tag -d "$tag" >&2
            fi
        done

        # Remove branches that don't point to any HEAD
        branchrefs=()
        eval "$(git for-each-ref --shell --format='branchrefs+=(%(refname))' refs/heads/)"

        for branchref in "${branchrefs[@]}"; do
            echo "Examining $branchref"
            rev="$(git rev-parse "$branchref")"
            echo "Has rev $rev"
            echo "List of revs: $revs"
            if [[ $revs != *" $rev"* ]]; then
                echo "Deleting $branchref"
                git update-ref -d "$branchref" >&2
            fi
        done

        echo "$revs" | while read rev; do
            echo "git branch b_$rev $rev"
            git branch b_$rev $rev
        done

        # Remove files that have timestamps or otherwise have non-deterministic
        # properties.
        rm -rf logs/ hooks/ index FETCH_HEAD ORIG_HEAD refs/remotes/origin/HEAD config

        # Do a full repack. Must run single-threaded, or else we lose determinism.
        git config pack.threads 1
        git repack -A -d -f
        rm -f config

        # Garbage collect unreferenced objects.
        git gc --prune=all
    )
done)

# Remove unneeded outputs
[[ ! -d $out/registry/src ]] || rm -rf $out/registry/src
[[ ! -d $out/git/checkouts ]] || rm -rf $out/git/checkouts